Sunday, October 18, 2026

The AI Psych Brief

🔬 Section 1

Psychiatric Research Article

Connectomic Determinants of Subcallosal Cingulate Deep Brain Stimulation for Depression

Song HN, Choi KS, Kopell BH, Figee M, Bruin J, Cha J, Riva-Posse P, Boutet A, Elias G, Germann J, Kennedy SH, Himes L, Gross RE, Lozano AM, Cheeran B, Mayberg HS · American Journal of Psychiatry · October 2026

Subcallosal cingulate (SCC) deep brain stimulation has had a mixed record in treatment-resistant depression: striking open-label responses, then disappointing sham-controlled results. This study pooled 131 patients from three independent SCC-DBS cohorts and reconstructed, for each person, the volume of tissue actually activated by their electrode settings.

Responders (at least 50% reduction in HAM-D) had a tighter, less variable spatial distribution of activated tissue than nonresponders, and stimulation mapping identified a focal cluster linked to improvement. Fiber-level analysis showed that what mattered was hitting specific white-matter bundles: failing to activate the left cingulum bundle and the medial uncinate fasciculus reduced the therapeutic effect.

💡 Why it matters

The lesson for the TRD referral conversation is that DBS outcomes depend on tract-level targeting, not just anatomical placement — which helps explain why earlier landmark-based trials underperformed and why connectome-guided implantation is now the standard to ask about. The data are retrospective and several authors have device-industry ties, so treat this as a targeting rationale, not new efficacy evidence.

Read the paper → doi:10.1176/appi.ajp.20250449

🩺 Section 2

General Medicine Article

Interventions to Improve Outpatient Follow-up for Hospitalized Patients With Alcohol Use Disorder: A Systematic Review

Bernstein EY, Hills-Dunlap K, Challapalli A, Bero L, Calcaterra SL, Edelman EJ, Kruse GR, Harnke B, Chopra V · Annals of Internal Medicine · September 2026

Hospital medicine researchers asked a question every consult-liaison psychiatrist faces: once a patient with alcohol use disorder is medically stabilized, what actually gets them to outpatient treatment? They reviewed 17 studies of 21 hospital-based interventions — 12 randomized trials (n = 1,917), 3 nonrandomized trials and 2 cohort studies.

Most interventions bundled several components, chiefly behavioral counseling and care coordination; only 14% included medication management. Just 8 of 21 interventions (38%) improved linkage to outpatient care, with effects in the RCTs ranging from an 8% absolute decrease to a 30% absolute increase. The ones that worked tended to combine coordination with AUD medications or social support. Certainty of evidence was low.

💡 Why it matters

Brief counseling and a referral slip are not enough on their own. When psychiatry is consulted on an inpatient with AUD, the most defensible move is to start naltrexone or acamprosate before discharge and attach a concrete, coordinated handoff — the components most often present in interventions that improved follow-up.

Read the paper → doi:10.7326/ANNALS-26-01312

💊 Section 3

Psychiatric Fact

Lithium during ECT: lower the level, don't stop the drug

Lithium plus ECT has a long case-report history of prolonged seizures and postictal delirium, and lithium also prolongs neuromuscular blockade from succinylcholine. The reflexive response — stop lithium for the ECT course — trades a modest procedural risk for a larger one: abrupt lithium discontinuation carries a well-documented risk of rebound mania and increased suicide risk, often in exactly the patient sick enough to need ECT.

The pragmatic middle path, supported by case series such as Dolenc and Rasmussen's Mayo review, is to keep lithium but run it lower: hold the evening dose before treatment (or the dose preceding each session) so the level at the time of stimulation is near trough, aim for the low end of the maintenance range, and tell the anesthesiologist so succinylcholine dosing and recovery are watched. If a patient develops marked postictal confusion, check the level before blaming the stimulus parameters.

🛋️ Section 4

Psychotherapy Teaching Pearl

Hate in the countertransference has to be owned before it stops leaking

Winnicott's 1949 paper argued that with very disturbed patients the analyst will come to hate the patient, and that this hate is often objective — a reasonable response to how the patient actually behaves — not merely a residue of the analyst's own conflicts. His technical point was that unacknowledged hate does not disappear; it gets enacted in small, deniable ways: shortened sessions, forgotten call-backs, interpretations sharpened into attacks, or overly solicitous niceness that is a reaction formation against the wish to be rid of the patient. Being able to tolerate and privately name the hatred keeps it in the therapist's mind instead of in the therapist's behavior, and paradoxically makes genuine concern possible again. The aim is not disclosure; it is knowing which of your interventions are carrying the feeling.

🗒️ Vignette

A PGY-3 notices she has started ending sessions with a man who berates her two minutes early, and that her interpretations have become clipped. In supervision she first describes feeling 'protective.' The supervisor asks what she would do if she let herself dislike him; she laughs and says she would cancel. Recognizing this, she keeps the full 45 minutes the next week and, when he again attacks, says calmly: 'You seem to need me to be someone who can't stand you — and then to prove I'll leave.' The move works because she is no longer secretly leaving, so the interpretation is not contradicted by her conduct.